A SERVICE CURVE APPROACH TO DEMAND RESPONSE Jean-Yves Le Boudec Dan-Cristian Tomozei 1 1.

A SERVICE CURVE APPROACH TO DEMAND RESPONSE

Jean-Yves Le BoudecDan-Cristian Tomozei

11

AgendaDemand Response

Service Curve ApproachUser Side Optimization

Operator Side Optimization

2

Demand Response

Some Demand can be delayed !DSO provides best effort service with statistical guarantees [Keshav and Rosenberg 2010]

3

Voltalis Bluepod switches off thermal load for 30 mn

PeakSaver cycles AC for 15mnProgrammable dishwasher

Price vs QuantityPeaksaver, Bluepod act by quantity control

DSO/Aggregator switches off appliance

Price control often proposed as alternative

Users save when price is high[Meyn 2010] : high volatility is an inherent feature of electricity markets

4

[Conejo et al, 2010]

Centralized vs Distributed ControlDirect control by DSO/Aggregator for air conditioning, dryers

Not scalable, does not adapt to diversity and flexibility

Appliance control should be done close to end-users

5

Price Based Approach

+ Distributed, flexible, user can interact

- Volatility, Reconciliation, Predictability

Quantity Based Approach

+ Predictable costs

- Centralized, inflexible, no user input

6

Service Curve Approach

+ Distributed, flexible, user can interact

+ Predictable costs

Definition of Service Curve Approach

1. Customer agrees to be throttled, with a bound

2. Fixed price per kWh3. Total load is controlled

7

DSO

Service curvecontract

Instant power

Control by DSO

Service curve

Example 1:Load

Switching

At most 30 mn of interruption total per dayOr reduction to for 60mn total per day

8

Example 2:Two Level

Control

Similar, but a minimum power is guaranteedBetter suited (than ex 1) when applied to an entire home /enterprise

9

The Maths of Two-Level

ControlThe constraint on is equivalent to

i.e. the allowed energy per window of time is lower bounded

10

User Side OptimizationUser can observe past signals and predict worst case futureSmart home controller can manage load accordingly

[LeBoudec Tomozei 2011]

11

Provider Side Optimization

Provider may send smooth signals

E.g. to many customers, for long periods of time

Or bursty signalsE.g. to selected customers, for shorter periods of time

Smooth signals are optimal for stationary but random loads, bursty signal are better for shaving peaks

12

EPFL Testbed

13

ConclusionsWe propose a service curve approach to demand response

DistributedApplies to total customer loadProvides large flxibility to providerProtects user from price uncertainty

14

[Le Boudec Tomezei 2011] Le Boudec J.Y. and Tomozei, D.C “Demand Response Using Service Curves”, EPFL-REPORT-168868, https://infoscience.epfl.ch/record/168868, 2011